One versus many independent assemblies of symbiotic nitrogen fixation in flowering plants

Doyle, Jeff J.; Ren, Jian; Pawlowski, Katharina; James, Euan K.; Gao, Yingzhi

doi:10.1038/s41467-025-60433-w

Download PDF

Perspective
Open access
Published: 19 June 2025

One versus many independent assemblies of symbiotic nitrogen fixation in flowering plants

Nature Communications volume 16, Article number: 5345 (2025) Cite this article

2388 Accesses
5 Altmetric
Metrics details

Subjects

Abstract

Some species of legumes and nine other flowering plant families form symbioses with bacteria that fix atmospheric nitrogen within specialized plant structures called nodules. How and how often nodulation symbiosis originated has implications for engineering symbiotic nitrogen fixation in non-legume crops. The prevailing hypothesis of a single origin with massive parallel losses has been challenged in a phylogenomic study favoring 16 origins and 10 losses. Nodulation has been assembled once or many times from existing processes (e.g., mycorrhizal symbiosis) and therefore almost nothing about it is truly novel. Because any feature of nodulation can be explained either as divergence from a common origin or as convergence in unrelated taxa, tests are needed that can distinguish whether assembly of homologous components has occurred uniquely or convergently. Much needs to be learned about nodulation symbioses across the proposed independent origins, especially involving the master nodulation transcription factor, Nodule Inception (NIN).

Comparative phylotranscriptomics reveals ancestral and derived root nodule symbiosis programmes

Article Open access 15 June 2023

Widespread Bradyrhizobium distribution of diverse Type III effectors that trigger legume nodulation in the absence of Nod factor

Article Open access 24 June 2023

Occurrence and diversity of stem nodulation in Aeschynomene and Sesbania legumes from wetlands of Madagascar

Article Open access 29 February 2024

Introduction

Atmospheric nitrogen is inaccessible to plants and must be converted into compounds such as ammonia. Plants have formed symbioses with nitrogen-fixing members of three major bacterial clades: cyanobacteria, Frankia, and diverse proteobacteria collectively termed “rhizobia.” In the latter two symbioses, flowering plants form nodules to house bacteria, most commonly on roots (root nodule symbiosis = RNS; nodulation). Nodulation of the important crop family, Leguminosae (Fabaceae; legumes, e.g., soybean and pea) by rhizobia has been studied intensively¹, not only as a fascinating symbiosis, but also with the goal of engineering nitrogen fixation in crops such as wheat and rice, with enormous potential benefits to agriculture and the environment². The feasibility of this goal has been linked to the evolvability of this trait and hence to whether nodulation has evolved once or many times. A landmark study by Kates et al.³ has added the latest chapter to this ongoing debate.

The first plant molecular phylogeny generated two competing origin hypotheses

Legumes and nine other flowering plant families with species engaging in RNS were widely scattered in pre-phylogenetic angiosperm higher-level classifications, so the default model was that nodulation had originated independently on numerous occasions from diverse genetic backgrounds. Surprisingly, however, the first large-scale chloroplast DNA-based phylogenies placed all nodulating species in a “nitrogen-fixing clade” within the eudicot rosid radiation⁴, a placement supported in subsequent phylogenies^3,5. Following Griesmann et al., we will refer to this as the “nitrogen-fixing nodulation clade” (NFNC); it has also been called “FaFaCuRo” after the two pairs of sister orders (Fagales and Fabales, which includes legumes; Cucurbitales and Rosales)^6,7 that together comprise the recently named fabid clade⁸.

But although nodulating lineages belong to a single clade, they are interspersed among many non-nodulating species (Fig. 1). To account for this complex distribution, Soltis et al. invoked a “predisposition” originating in the most recent common ancestor (MRCA) of the NFNC, and proposed two competing hypotheses⁴:

**Fig. 1: Multiple origins and losses of nodulation hypothesized by Kates et al.³ in the four orders of the NFNC.**

“…if the underlying predisposition to engage in symbiotic nitrogen fixation stems from a single, common origin, members of this clade that do not share this symbiosis must have lost the ability to form such associations. Alternatively, the ancestor of the nitrogen-fixing clade may have evolved the genetic components that would ultimately permit the evolution of symbiotic nodular nitrogen fixation. Following the establishment of these conditions, the necessary genetic background was present to allow parallel, recurrent evolution of symbiotic nitrogen fixation in the subsequent diversification of this clade. Both hypotheses call for mutations that established the symbiotic association with nitrogen-fixing bacteria, or at least the basis for this symbiosis, in the ancestor of the nitrogen-fixing clade [italics and bold face added].”

Clearly, for Soltis et al.⁴, “predisposition” is a feature shared by two competing hypotheses, in both cases providing “the necessary genetic background” for nodulation to evolve in the NFNC. Subsequently, however, “predisposition” came to refer exclusively to an unknown trait that was present in the NFNC ancestor but only much later enabled “parallel, recurrent evolution of symbiotic nitrogen fixation”⁴. We will refer to this hypothesis of multiple origins of a nodulation syndrome as “MUL” and the alternative hypothesis of the full nodulation symbiosis synnovation⁹ originating in a “single common origin⁴” as SNG (Fig.1). We will follow Werner et al.¹⁰ in referring to this as the “precursor trait,” using “predisposition” in its original sense to refer to both hypotheses. That Soltis et al.⁴ did not favor either hypothesis is seen in their conclusion that “nodulating and non-nodulating members of this nitrogen-fixing clade should be examined to ascertain whether recurrent losses or recurrent gains of nitrogen-fixing ability have occurred.”

For nearly a quarter of a century, as increasing numbers of species were phenotyped for nodulation (Box 1), the single origin hypothesis was ignored, with debate centering on how many independent origins had occurred. This period saw the advent of phylogenomics, including more sophisticated approaches to estimating the number of origins of nodulation¹⁰ that presaged the Kates et al. study (Box 2). Much attention focused on the unsuccessful search for the predisposition/precursor, but some attention was paid to the question of how to determine whether nodules in different groups were homologous—i.e., derived from the same origin¹¹.

Box 1 Phenotyping nodulators

“Nodulation” as defined here requires that macrosymbionts (plants) house microsymbionts (Frankia or rhizobia) in plant-produced nodules, within which the bacteria fix atmospheric N₂ into forms metabolically accessible to plants and in turn are provided with carbon derived from photosynthesis. Thus, nodulation is a binary trait: A species either has nodules or lacks them.

Effective nitrogen-fixing nodules are a facultative trait, and there are conditions where nodulation is negatively affected; e.g., salt stress will reduce nodulation. Nevertheless, in natural ecosystems (i.e., ecosystems without N fertilizer), nodulators are normally found nodulated, and this is an easily detected trait. However, there can be problems with false negatives and false positives. In general, if one member of a genus is a nodulator, so are the other species. If there are ostensible exceptions, how to prove a negative? When a plant is not nodulated in the field, this may be due to a lack of compatible microsymbionts. If it cannot be made to nodulate in the greenhouse, the microsymbionts used may not be compatible.

To account for the polyphyly of the large mimosoid legume genus Acacia sensu lato seven distinct segregate genera (several of them based on resurrected subgeneric/sectional names) are now recognized⁹³. One of these segregate genera, Senegalia, is itself non-monophyletic, and most likely comprises at least two genera⁹³, both of which contain nodulators and non-nodulators. These do not segregate according to the two groups within Senegalia, i.e., both of the main subclades of Senegalia species have Nod+ and Nod- species. The former are long recognized, and there are numerous reports about the latter⁹⁴. Genome sequencing of four Nod- and Nod+ species selected from each of the main subclades of Senegalia is currently in progress to determine the genetic basis of these contrasting nodulation phenotypes.

The North American actinorhizal species Dryas drummondii is the only nodulator in the Dryas genus. In the field, nodulation of D. drummondii is not consistent; there will be one nodulated root system among 20 to 30 dug out from a patch of D. drummondii (J. Dawson, personal communication). However, when using seedlings in the greenhouse and soil collected from below a nodulated plant, large amounts of nodules will be found. Is the Nod- phenotype of the other Dryas species reliable? D. octopetala is far more widespread than D. drummondii and was never found nodulated; since D. drummondii forms arbuscular mycorrhiza while D. octopetala does not form arbuscular mycorrhiza but instead ectomycorrhiza, it is generally accepted that at least this species cannot nodulate. Ultimately, though, also here, plant genome sequencing will have to provide the final answer.

Regarding false positives—root nodules not induced by nitrogen-fixing bacteria—nodules not containing nitrogen-fixing symbionts can occur in nodulators. One example would be the myconodules of Alnus incana, induced by Penicillium nodositatum⁹⁵on this plant, which will also form nitrogen-fixing nodules. Furthermore, members of the conifer families Podocarpaceae, Araucariaceae and Phyllocladaceae, as well as some members of the legume subfamilies Caesalpinioideae and Detarioideae⁹⁶ plus a single actinorhizal genus, Gymnostoma (Casuarinaceae), can form so-called myconodules, short lateral roots with determinate growth and strong arbuscular mycorrhizal colonization in the expanded cortex⁹⁷. Recently, Rahimlou et al.⁹⁸ described root nodules of Tribulus terrestris (Zygophyllaceae) and Roystonea regia (Arecaceae), which looked like young actinorhizal nodules and contained a complex microbiome including, in the case of T. terrestris, Sinorhizobium meliloti at very low density, and no Frankia strains in either plant, i.e., the nodules did not harbor a substantial population of symbiotic diazotrophic bacteria. Thus, false positives could always be identified by metagenome analysis, or based on the fact that they do not provide growth support via N supply.

Box 2 Many gains, but how many: from implicit parsimony to hidden rate models

Given the complexity of RNS, the evolutionary gain of nodules has seemed much more difficult than loss^1,11, but because a single origin of nodulation in the common ancestor of all nodulating NFNC species requires so many independent losses, SNG was widely ignored^{5,10,12,20,43}. MUL was more appealing as an explanation for phylogenetic results based on implicit invocation of the parsimony principle⁵ and its assumption of convergent gains of non-homologous nodules was also supported by the many differences in nodule biology across the four NFNC orders suggested (Fig. 1)⁴⁸. Debate centered on speculation about how many independent gains seemed reasonable, given that loss was still more readily explainable than gain, and that re-gain of nodulation once lost was highly improbable. Thus, nodulation origin was implicitly a weighted Dollo parsimony process⁹⁹, lacking any theoretical justification for assigning relative weights to loss vs. gain—a general problem in systematics.

Beaulieu et al.³² noted that simple models of character evolution, such as parsimony, though useful for small groups of closely related taxa, are inadequate in large clades because rates of change would almost certainly vary across the phylogeny, leading to incorrect mapping of characters onto phylogenies. They adapted methods for dealing with site-specific rate heterogeneity in DNA sequence data (e.g., the higher rate of transition vs. transversion substitutions) for use with binary non-molecular characters, and created a hidden rate model (HRM) that “can be used to allow different rate classes to be treated as ‘hidden’ states in reconstructing ancestral character states.” The incorporation of hidden states reflects the reality that the factors leading to change in the propensity of a character to originate (or be lost) may have nothing to do with the character itself. To use a molecular example, a decrease in generation time can lead to an increase in substitution rate observed across the entire genome. Such hidden states can be anything not accounted for explicitly in the model, not only traits of the taxa under investigation, but ecology or interactions with other organisms³². Therefore, it is not clear how—or even if—they can be tested.

Werner et al.¹⁰ applied the HRM approach to a 32,000 taxon plant phylogeny of Zanne et al.¹⁰⁰, using a comprehensive database of nodulation occurrence built on the lifetime of work by co-author Janet Sprent, the pre-eminent authority on the distribution of nodulation. Their best models supported MUL, with the independent evolution of the full nodulation symbiosis occurring in around eight different NFNC lineages, but only in lineages that retained a precursor trait. This precursor originated only once, in the NFNC ancestor, and persisted in some lineages for over 30 MY before enabling independent origins of nodulation. The precursor trait is relatively easily lost and is absent from numerous lineages throughout the NFNC, but loss of nodulation occurred only around 10 times.

This pioneering study thus supported the community’s working hypothesis of multiple origins from an enabling precursor trait. However, it did not resolve where and when nodulation had arisen^10,101, largely due to serious flaws in the phylogenetic tree^100,102. More recently, Zhao et al.⁶¹ generated a densely sampled legume phylogeny; they used the same HRM approach³² to illustrate scenarios of 1–3 nodulation origins in Fabaceae without indicating a preference.

The unexpected “resurrection” of the single origin hypothesis

Distinguishing convergence from homology in nodulation is challenging¹¹. In a short essay focusing on legumes, Doyle¹² suggested that, rather than concentrating exclusively on testing homology of nodules, the problem of the number of independent origins might be addressed by studying non-nodulating species, distinguishing primary absence of nodulation from secondary loss. Species that had once been able to nodulate might preserve “fossil” evidence in their genomes. A similar approach had been used by Delaux et al.¹³ to detect unidentified genes involved in mycorrhizal symbiosis by comparing genomes of mycorrhizal plants with those of non-mycorrhizal Brassicales.

In 2018, two groups (one co-led by Pierre-Marc Delaux) reported that in nearly all non-nodulating NFNC taxa with sequenced genomes, a few key nodulation genes found in all nodulating species were either absent or were pseudogenes, and interpreted this as evidence for SNG^6,14. The “resurrected” single gain/massive parallel loss hypothesis as expounded by van Velzen et al.¹⁵ proposes that a nodulation synnovation originated in the NFNC ancestor and has been maintained in all nodulating lineages (Fig. 1; SNG)^6,14. The predisposition that is unique to the NFNC is part of the nodulation synnovation itself⁴, so all features of nodulation would only need to be assembled once rather than many times as in MUL, regardless of whether they evolved de novo or, more likely, were recruited from pre-existing functions.

A critical and under-appreciated feature of SNG is that although it is commonly described as the “single gain/massive parallel loss” hypothesis, it does not require any “loss” of nodulation to account for the phylogenetically scattered distribution of lineages that once nodulated but no longer do so. That is because nodulation is an optional, inducible, plastic phenotype, and whenever the cost of nodulation exceeds the benefit of symbiotically obtained nitrogen (as in a soybean field treated with exogenous fertilizer) plants do not invest in making nodules^16,17,18,19. In the millions of years since the common ancestor of the NFNC, photosynthetic carbon has become increasingly expensive as atmospheric CO₂ declined^15,20,21. This had major effects on the evolution of plant, animal, and bacterial lineages and the ecosystems they inhabited^22,23,24,25; critically, microsymbionts might no longer be present in the soil owing to radical environmental change (e.g., flooding, drought, soil pH and/or biochemical shifts)²⁶. Although Li et al. had proposed that decrease in atmospheric CO₂ could drive multiple origins of nodulation⁵, this would require the precursor trait to evolve at a period of relatively low CO₂ cost, with individual taxa “waiting” millions of years before evolving nodulation when the C/N cost/benefit ratio was much less favorable for nodulation⁵. It seems much more likely that the decreasing benefit of nodulation would lead to its cessation.

In SNG, cessation of nodulation would be inevitable, not merely likely, whenever carbon cost exceeded nitrogen benefit, but that point would vary for each individual plant depending on the interactions between its genotype and that of its bacterial partner, and of both with the environment (G_p × G_b × E)^27,28. Cessation is not loss, as is demonstrated by the plasticity of nodulation in extant species. However, similar to the decay of vision-related characters in another massively parallel phenomenon, blindness in cave-dwelling organisms²⁹, unused key nodulation components would no longer be preserved by purifying (negative) selection, and would accumulate mutations stochastically that over time would inactivate them. Nodulation genes not under selection for other functions would become pseudogenes and could be lost entirely; regulatory elements, domains, or specific nucleotides involved in N-fixing symbiosis could be modified or lost from genes retained due to their non-nodulation functions. Lineages could preserve variable amounts of evidence of having once nodulated, as observed in the genomes of non-nodulating NFNC taxa^6,14. Loss of nodulation-specific features could possibly be driven by selection³⁰. Many members of the nodulation community were authors on either Griesmann et al.⁶ or van Velzen et al.¹⁴, so it is not surprising that SNG became the prevalent working model for nodulation evolution, though alternatives continued to be proposed³¹.

Resurgence of the multiple origins hypothesis

Kates et al.³ have now generated a robust, updated multiple origins hypothesis. This new formulation of MUL hearkens back to an earlier modeling study (Box 2) but with a bespoke phylogenomic analysis of ~13,000 mostly NFNC species based on 100 low copy nuclear genes; an augmented, comprehensive, expertly curated database of nodulation; and an updated version of the hidden rates model (HRM) approach^32,33. Their best model identifies 16 independent origins of nodulation across the NFNC, with only 10 losses, all occurring within the legume family. As Kates et al.³ note, “The overall pattern of gain and loss resembles that recovered previously¹⁰, with more losses and gains reported here due in large part to our improved sampling of clades with high rates of RNS transition” (Fig. 1). Kates et al. counter expected criticism by noting that convergence is rampant in evolutionary history; they point in particular to C₄/CAM photosynthesis, which has similarities but also key differences from nodulation (Box 3).

MUL sensu Kates et al. involves several different transitions between both observed and hidden states. First, an unidentified precursor trait, necessary but not sufficient for nodulation, evolves as in Soltis et al.⁴, but deeper in the rosid clade, earlier than the NFNC MRCA. This precursor state can either be lost (revert to the non-precursor state, e.g., in rosid lineages outside the NFNC), eliminating any chance of nodulation evolving, or, more rarely, can evolve to an intermediary hidden state, increasing tenfold the probability of nodulation evolving. The intermediary state itself can evolve either to (in descending order of probability) the non-precursor state, a transient nodulation symbiosis state (R1), or the precursor state. R1 can evolve to a stable nodulation symbiosis (R3) or to a second transient nodulation state (R2), which almost always reverts to the precursor state but also has a low probability of evolving a stable R3 nodulation symbiosis. These unknown transient states need not be homologous and may have nothing to do with nodulation (Box 2; e.g., R1 in one lineage could be an ecological shift, whereas R1 in another lineage could be a gene duplication). Kates et al.³ concluded that although genetic components leading to nodulation are easily lost, nodulation once gained is highly stable.

Among the most compelling features of the Kates et al.³ study is the rigorous testing of models with parameters strongly favoring SNG. Despite allowing the probability of loss to exceed that of gain by over 60-fold, a single origin of nodulation was always overwhelmingly rejected. It should be recalled, however, that loss of nodulation is not an essential part of SNG; what is important is cessation, which is inevitable and presumably incurs no cost at all. “Loss” is an inference; the observation is absence, and dealing with absence characters in phylogenetics is notoriously difficult^34,35. We doubt that it is possible to parameterize an optional trait that ceases automatically when no longer beneficial, and/or when the microsymbiont disappears or becomes parasitic, Thus, we do not agree with the contention of Kates et al.³ that their results “definitively” identify the ancestor of the NFNC as non-nodulating.

Despite our disagreement, however, we believe that Kates et al.³ should be welcomed by the nodulation community as a major improvement over intuitive, parsimony-based mapping of nodulation on small phylogenies, and although sampling is still not exhaustive (e.g., some key legume genera are missing and there are mistakes in identification of terminal legume taxa), it is a great improvement over previous modeling based on flawed phylogenies¹⁰. It thus constitutes the best available MUL hypothesis for rigorous testing, and can guide innovative exploration of RNS evolution, such as conducted by Battenberg et al.³⁶, who found signatures of positive selection associated with nodulation in several lineages. Kates et al.³ acknowledge that details of MUL will almost certainly change as additional taxa are sampled. Throughout their paper, they highlight the need for more information on the molecular and developmental biology of nodulation, particularly in under-studied lineages—which is most of the NFNC except for a few model (mostly legume) taxa—and for a more sophisticated definition of nodulation itself.

Box 3 Analogies of RNS with the evolution of C4 photosynthesis and eyes

The evolution of C4 and CAM photosynthesis is probably the best-documented example of convergence in plants¹⁰³. Of these two adaptive modifications of standard C3 photosynthesis, the evolution of C4 is far better understood, largely due to relatively unambiguous physiological, anatomical, and morphological criteria available for distinguishing C4 from C3 plants¹⁰⁴. Much like the situation in RNS from 1995 to 2018 (Box 2), convergence has been assumed, and as observations were amassed and optimized on increasingly saturated phylogenies, by 2011, 62 origins of C4 were hypothesized¹⁰⁵. As in RNS, the C4 syndrome has been assembled from photosynthetic and core metabolic modules exapted from C3 photosynthesis^88,106, which are homologous across different origins; as in RNS, there is very little that is unique about C4. There is another parallel, as Christin et al. had noted presciently in 2010¹⁰⁷

“Establishing the relationships between C₃ and C₄ taxa is thus not sufficient to assess the number of C₄ origins and reversions. These issues must instead be tackled by establishing the homology of the genetic determinism and the morphological and biochemical nature of the C₄ syndromes exhibited by different C₄ groups. Conversely, reversals can only be proven by detecting genetic traces of C₄ ancestry in present-day C₃ plants, demonstrating that the C₃-specific characters of putative revertants are homoplasious compared to the C₃ trait of the ancestors.”

The progress being made in dissecting homologies across C4 species is an area that offers inspiration to those studying the evolution of RNS. Of particular note are two recent studies conducted at the cell/nucleus level. Swift et al.¹⁰⁶ show that photosynthetic genes expressed in the mesophyll of C3 rice have come to be expressed in bundle sheath cells of C4 Sorghum. This switch in the ___location of expression of photosynthesis genes has involved swapping of cis-regulatory elements conferring cell type-specific binding of key transcription factors—the repeated co-opting (exaptation) of an existing trans-code through pervasive rewiring in cis. Mendieta et al.⁸⁸ assayed chromatin accessibility in rice and four grass species, each thought to have evolved C4 photosynthesis independently. Their results reveal how dynamic C4 evolution is in these species. In some cases, independent origins of C4 have involved the recruitment of paralogous gene copies, whereas in other cases the same (orthologous) gene copy has been recruited in parallel by different C4 species; but they provide an example of how rapid evolutionary divergence from the common C4 ancestor of Zea and Sorghum has led, in only ~13 MY, to paralogous copies being expressed. This same dynamism was discussed many years ago as complicating the use of gene expression criteria for inferring RNS homologies¹¹.

Although RNS and C4 have parallels, they differ in ways that limit the utility of the analogy. Most notably, C4 plants did not have to invent photosynthesis, only modify it. The focal anatomical feature of most C4 plants is the bundle sheath, also found in C3 plants, whose transformation as part of a convergent suite of anatomical characters appears to have been a requirement for the multiple origins of C4¹⁰⁴. In contrast, the evolution of RNS required the origin of the nodule, an organ of unknown homology³⁹. Whereas the bundle sheath cell type is found in C3 plants, in RNS, host cells infected with bacteria represent novel cell types that are the key units of the symbiosis. Moreover, although the DOF TF family plays a key role in shifting cell type expression in Sorghum⁸⁸, in RNS, a single TF, NIN, coordinates modules not known to be associated functionally outside of RNS.

N-fixing symbiosis can be analogized with the evolution of another novel structure, the eye. The non-homologous eyes of different phyla (mollusks, vertebrates, etc.) are analogous to non-homologous N-fixing symbioses in plants (cycads, ferns, legumes), and have something to offer those studying RNS, particularly proponents of MUL as illustrations of how complex novel structures originate convergently. For proponents of SNG, RNS, as one origin of N-fixing symbiosis, could be compared with the evolution of one particular type of eye, such as the vertebrate camera-type eye, which has evolved only once, and whose dedicated components decay when the visual apparatus ceases to be used, as in cave vertebrates. The convergent process of decay of visual components as in cave organisms from other phyla is also a key part of the SNG hypothesis.

Deconstructing “nodulation”: testing for single vs. multiple assembly of homologous parts

It has long been accepted that the evolution of the complex association of integrated characters that comprise nodulation involved the co-option, rather than the invention, of individual genes, and the recruitment or rewiring of existing gene regulatory networks (GRNs)^{3,4,11,12,37,38,39,40,41,42}. Homology of individual nodulation modules in all nodulating lineages is a given in SNG, so non-homology of genes or modules would support MUL. For example, if genes recruited for similar RNS functions in two nodulating taxa are non-homologous, or paralogous rather than orthologous, then the assembly was more likely independent, making nodulation non-homologous¹¹.

Independent origins of nodulation likely would involve convergent or parallel recruitment of the same genes or GRNs, with the result that nodulation would be non-homologous, but some or all of its components could be homologous^{3,10,11,12,43}. Such “deep homology”⁴⁴ confounds simple definitions of homology for complex traits because it can be invoked to discount any amount of similarity^45,46. This has consequences for inferring common RNS origin based on greater than random transcriptomic similarity in nodules of taxa from different NFNC orders^14,36. For example, Libourel et al.⁴⁷ reconstructed a core ancestral nodulation transcriptome of over 750 orthogroups shared across three orders of nodulating NFNC species, which was much higher than random expectation, and concluded that “RNS in the most recent common ancestor looked very similar to RNS in extant species.” But deep homology stems from the non-random recruitment of homologous GRNs, in parallel, for use in functionally similar syndromes, blunting such conclusions.

The eponymous organs of NFNC members are all called “nodules” because they meet the criterion of similarity in form and position that is the basis for the most common evolutionary/phylogenetic definition of homology as “similarity due to common descent.” However, the structural and biochemical diversity of nodules across the NFNC⁴⁸ has favored non-homology and thus MUL. Under SNG, this diversity is explained by >100 MY of independent evolution since the divergence of modern lineages from the NFNC ancestor. We note that the diversity of nodules is matched, if not surpassed, by the flowers of these same NFNC species (Fig. 1), yet it has not been suggested that flowers are non-homologous in the clade. And although Kates et al.³ cite Liu et al.⁴⁸ for “limited homology” of nodulation across the NFNC, much of that paper details differences among the homologous nodules of papilionoid legumes, which arguably differ phenotypically more from one another than some of them do from putatively non-homologous (under MUL) nodules of caesalpinioid legumes (Fig. 1)⁴⁹. Moreover, looks can be deceiving. Shen et al.⁵⁰ constructed an actinorhizal plant nodule fate map based on Parasponia andersonii and Alnus glutinosa, which revealed that the developmental processes of actinorhizal plant nodules have greater similarity to those of legume nodules than previously appreciated. Then, by mutating a single transcription factor (NOOT) in Medicago, they produced a homeotic change from legume-type to actinorhizal-type nodule anatomy.

What is truly novel about nodulation is the recruitment and assembly of a set of modular components into a functional nitrogen-fixing symbiosis with bacteria housed in plant-produced structures (Fig. 2a). In exploring how RNS might be engineered in crop plants, Jhu and Oldroyd⁵¹ provide an excellent description of these components and discuss how they might have been assembled at the origin of nodulation. They emphasize the longstanding paradigm that much of RNS has been recruited from the older mycorrhizal symbiosis, itself “appropriated” from plant defense responses⁵². Both symbioses begin with the Common Symbiosis Signaling Pathway (CSSP), but beyond that, there has been massive transcriptomic reprogramming leading to nodulation³¹, key to which is a single transcription factor (TF), Nodule Inception (NIN). The many roles of this master regulator of nodulation^53,54,55,56 afford an opportunity for breaking “nodulation” into its constituent parts and asking when and where these were brought together—once or multiple times (Fig. 3)—in rosid phylogeny. As Geurts and Huisman⁵⁷ note, understanding how NIN has been adapted to function in nodulation is critical for understanding the evolution of RNS.

**Fig. 2: Stepwise recruitment of features of nodulation.**

**Fig. 3: Relationships between processes shared between nodulation and other functions under the two competing hypotheses for legumes (*Medicago*) and Betulaceae (*Alnus*).**

Box 4 Polyploidy and the evolution of nodulation

Whole genome duplication (WGD, polyploidy) has shaped the genomes of all flowering plant lineages, with rosid genomes averaging 3–4 WGD events in their histories¹⁰⁸. Polyploidy is viewed as a major driver of phenotypic novelty¹⁰⁹, so it is not surprising that, since the discovery of polyploidy in papilionoid legumes, there has been speculation that genome duplication has led to or “enhanced” nodulation there, and perhaps in other legume lineages as well^{61,101,110,111,112,113,114}. It is rarely possible to compare modern polyploid genomes formed 30–100 MYA with their now-extinct diploid progenitors, but studies of much more recent polyploidy (<1 MY) on nodulation have shown that polyploidy can indeed enhance nodulation^115,116.

The relationship between multiple legume polyploidy events and the phylogenetic distribution of nodulation, already complex¹⁰¹, has more recently been further complicated by the finding that allopolyploidy may have been involved in early legume evolution¹¹⁷, adding hybridity to factors potentially influencing the origin or enhancement of nodulation. Polyploidy occurs throughout the NFNC, characterizing lineages from orders (e.g., Cucurbitales¹¹⁸) to individual species. The eventual (and inevitable) saturation of the NFNC with fully sequenced genomes will provide the information needed to correlate the distribution of polyploidy with the origin, loss, or enhancement of nodulation under MUL and SNG.

Under SNG, although WGD early in eudicot history would factor into the evolution of nodulation, there are no WGD events known from the immediate ancestor of the NFNC. A role for polyploidy in the origin(s) of nodulation is easier to envision under MUL because there are many opportunities for polyploidy to have affected individual components of nodulation as they evolved independently in different lineages over millions of years, beginning with the precursor, which in MUL arises prior to the NFNC ancestor. It is also possible that some of the hidden states of Kates et al.³ could turn out to be WGD events. Cessation of nodulation should lead to the decay of any paralogues that had specialized solely for nodulation, regardless of how many homoeologous copies have been created in highly polyploid lineages. The degree of polyploidy therefore does not favor either MUL or SNG.

Polyploidy is cyclical¹¹⁹, with genome multiplication followed by processes lumped under the catchall term “diploidization.” A modeling study by Parshuram et al.¹²⁰ revealed the complexity of interactions among polyploidy, diploidization, and nodulation on the ecology of legumes. Unfortunately, they used the same flawed phylogeny as Werner et al.¹²; in their best models, the ancestral legume was diploid and, as in Kates et al., it was non-symbiotic. The dynamic polyploid/diploidization cycle involves loss of many duplicated genes; retention of duplicates often involves changes in function or expression pattern. These processes shape the evolution of gene families, and could thus underlie the origin or enhancement of nodulation^121,122,123.

A brief history of NIN

The NIN-like protein (NLP) transcription factor gene family is plant-specific, with members belonging to three major groups in seed plants, playing key roles in the plant nitrate response^58,59. Group 3 is duplicated in dicots, with one paralog comprising the NIN clade, within which there has been further amplification in various taxa (e.g., three lineages in Brassicaceae⁹). The three Arabidopsis paralogues (NLP1-3) play similar roles in nitrate sensing and share this with at least one member of the other Group 3 subclade, NLP7⁶⁰, making it likely that this is the ancestral role of both subclades.

It has been suggested that the loss of this role, through deletions in an otherwise conserved nitrate-responsive ___domain (NRD), was critical for the evolution of RNS⁵³; however, NIN proteins of nodulating species outside papilionoid legumes retain much or all of the NRD^9,61. Regardless of whether the NRD is functional in these NIN orthologues, the loss or inactivation of NIN in many non-nodulating NFNC lineages suggests that it is no longer essential for nitrate-sensing.

The co-option of mycorrhizal signaling through NIN (and Nod-factor receptor genes) required changes not only to the NIN protein but also to the NIN promoter region⁵⁷. The CYCLOPS-responsive element (CYC-RE) links NIN to the most downstream CSSP TF, shared by RNS and mycorrhizal symbiosis⁶². Though not essential, it is important for infection thread formation in legumes⁶³. The CYC-binding motif is part of a larger Predisposition-Associated Cis-regulatory Element (PACE) that Cathebras et al.⁶⁴ identify as the nodulation predisposition/precursor trait, being found only in the NFNC, both in nodulating species and in non-nodulators retaining NIN. They also showed that PACE drives NIN expression during cortical infection thread development in Lotus and that, despite sequence differences, PACEs from the other three NFNC orders, including both nodulating and non-nodulating species, could rescue a Lotus nin mutant⁶⁴. Moreover, the addition of PACE to a tomato NIN gene promoter also reactivated RNS in the Lotus mutant⁶⁴.

Cathebras et al.⁶⁴ identified intact NIN open reading frames in 10 of 28 non-nodulating NFNC species, but PACE was absent from five of these; from this, along with the presence of PACE in NIN of all 31 nodulating species surveyed, they concluded that PACE was functional in the NFNC MRCA, and was generally lost when nodulation ceased. However, although PACE was not reported in poplar, the CYC-RE occurs in a poplar Nod-factor-inducible NLP gene involved⁶⁵, like NIN⁵⁴ and other NLPs⁵⁸, in lateral root development. This “cast doubt upon the theory that the essential innovation in the evolution of nodulation was the gain of regulation of NIN by the common symbiosis pathway”⁶⁵.

According to Shen and Feng⁵⁶, “NIN controls all aspects of symbiotic nodulation in legumes: rhizobial infection, nodule organogenesis, transition to nitrogen fixation, and regulation of nodule number in legumes and actinorhizal plants.” This sweeping statement is based almost exclusively on model papilionoid legumes. Although NIN is known to be essential for nodulation in Casuarina^66,67 and Parasponia⁶⁸, and is implicated in RNS of Ceanothus, Datisca, Alnus, and Purshia^9,36,69, the role of NIN is completely unknown in the remaining nine Kates et al.³ RNS origins.

In legumes, NIN interacts directly with many genes belonging to the various modules recruited in the origin or subsequent evolution of RNS^56,70. Any of these genes that did not previously interact with NIN would need to acquire NIN-binding sites in their promoters either once (SNG) or convergently (MUL), as appears to have occurred in the origins of C4 photosynthesis in grasses (Box 3). Much remains to be learned about promoter structure in NIN-responsive genes; an example is the complex dual promoter in Lotus leghemoglobin genes that binds NIN and NLP2, which belong to different NLP Group 3 subclades⁷¹. Yu et al. describe NIN-binding sites in the promoters of an NFNC-specific gene, Nodules with Activated Defense (NAD1)⁷². The gene and these “nodulation-associated cis-regulatory elements” (NACEs) are found in both nodulating and non-nodulating NFNC members. NACEs have sequence similarity to NIN-binding sites in known NIN targets^{41,54,71,73,74,75,76,77}. Nodulation in Lotus mutants lacking Nad1 NACE sites was rescued by introducing NACE sequences from Alnus, Parasponia, Casuarina, and Datisca, demonstrating the conserved nature of these sites⁷²; functionality was not tested for NACEs from two non-nodulating NFNC species.

Questions regarding nodulation, emphasizing the role of NIN

SNG is founded on the thesis that NIN was lost or pseudogenized in many non-nodulators because its role was solely associated with RNS; the retention of functional NIN in some lineages is consistent with neutral expectations (loss is random but not universal) or with the acquisition of functions not occurring in lineages where NIN is dispensable. In contrast, Kates et al.³ suggest that “essential RNS-related genes could be recruited without an RNS phenotype and then become dispensable and thus fragmented or lost in lineages that do not also eventually gain RNS”; they point to the presence of intact NIN in non-nodulators such as Ziziphus (Rhamnaceae) as support for MUL. Understanding the occurrence of NIN and characterizing its function in phylogenetically diverse nodulating and non-nodulating taxa is essential to testing the competing hypotheses.

What is the distribution of intact, functional, and pseudogenized NIN across the NFNC?

Plant gene families are constantly evolving by duplication, producing novel functions and expression patterns but more commonly leading to pseudogenization and loss⁷⁸. Polyploidy duplicates all genes simultaneously and has been implicated in the evolution of nodulation (Box 4). Soybean has retained all four homoeologous NIN copies from the two most recent rounds of polyploidy that characterize its genome, and these show a complex pattern of expression at the tissue and single cell level^79,80. This degree of differentiation among paralogous copies suggests that generalizations concerning “NIN” evolution and expression, particularly those based on transcriptomic studies of bulk tissues, may need to be re-evaluated. The condition of NIN is unknown for most NFNC, including key legume lineages. Comprehensive sampling of NIN could enable application of an approach similar to that of Kates et al.³ to estimate the number of losses and the correlation of NIN presence with nodulation. Minimum ages of NIN inactivation could be tested in lineages retaining NIN pseudogenes by identifying pseudogenizing mutations shared among species. The complexity of the NLP family, and of NIN, makes synteny information critical for the correct identification of NIN and would enable the detection of partially decayed pseudogenes.

How does NIN function in nodulation across the NFNC?

Is NIN constitutively localized to the nucleus in species other than the few papilionoid legume species in which this has been shown⁵³? How many of the modular functions of nodulation controlled by NIN in model papilionoids are shared and regulated by NIN in other lineages, including crown papilionoids, where nodulation is homologous? For shared modules, are the same GRNs involved? Liu et al.⁷⁰ identified 145 Medicago genes with which NIN interacts directly; how many of these genes interact with NIN in other nodulators, both within papilionoid legumes and in other putatively independent nodulating lineages? Are genes that interact directly with NIN in different nodulating lineages paralogous, which would support MUL¹¹? Identification of genes interacting with NIN, for example, by reciprocal DAP-seq⁸¹ experiments involving multiple Kates et al.³ MUL lineages, would be informative.

What is the structure of the regulatory regions of genes targeted by NIN across the NFNC?

The “cis-regulatory code” is poorly understood⁸², but tools are being developed to “crack” it, even in plants⁸³. For nodulation, the critical evolutionary question is how NIN recruited and brought together new targets and itself came under the regulation of the CSSP⁴¹; regulatory sequences are key to understanding this “rewiring”^52,84. Because nearly all genes involved in nodulation presumably have non-nodulation functions, genes recruited to function under the control of NIN possess binding sites not only for NIN but for transcription factors involved in their original, conserved functions^6,9,14,57. TF binding sites are short, conserved motifs often occurring within cis-regulatory modules in which the order of binding sites for different TFs may be conserved⁸⁵. If promoters containing NIN-binding sites could be identified and related to target genes by reciprocal binding experiments involving numerous MUL lineages, these could be compared phylogenetically across lineages for evidence of potential homology of NIN-binding sites (Fig. 4).

**Fig. 4: Homologous and non-homologous gene recruitment inferred from TF binding sites.**

Pereira et al.⁸⁶ and Zhang et al.⁹ used genome-wide alignments of NFNC species to identify thousands of conserved noncoding sequences/elements (CNS/CNE), many associated with RNS, with some validation by Pereira et al.⁸⁶. These provide a rich source of candidates for further study. Kates et al.³ MUL provides guidance for assessing the presence/absence of CNEs in the context of RNS evolution. The work of Yu et al.⁷² on NIN-binding sites, combining phylogenetic and functional studies, provides a template for investigations of promoters of other genes targeted by NIN, such as NFY and RPG^73,75,87. Recent studies of accessible chromatin in the evolution of C4 photosynthesis in grasses⁸⁸ provide a template for studying recruitment in building complex phenotypes.

What roles does NIN have apart from nodulation?

The Kates et al.³ contention that “essential RNS-related genes” such as NIN could be “recruited without an RNS phenotype” requires that NIN have some role other than RNS and that it retain functionality for millions of years before either becoming a key player in RNS independently in 16 NFNC lineages or being lost in many lineages that did not evolve RNS. There is no known role for NFNC orthologues of NIN other than RNS, but it is possible that NIN could conserve the ancestral nitrate-sensing function of Arabidopsis NIN-clade orthologues in NFNC taxa whose NIN retains most or all of the NRD. This includes not only non-legumes (e.g., Quillaja⁸⁹, sister to Fabaceae in Fabales), but also sampled caesalpinioid legumes^9,61. This should be tested, particularly in non-nodulators.

What is the fate of nodulation genes when nodulation ceases?

SNG and MUL sensu Kates et al.³ agree that 10 formerly nodulating lineages no longer engage in RNS. In addition to these, because Dryas drummondii is the only nodulating Dryas species, there is an additional recent “loss” of nodulation in the genus⁹⁰ not mentioned by Kates et al.³. How have promoters of direct NIN target genes evolved in these taxa, and how has NIN and its promoter region evolved? This information could provide a baseline for distinguishing the primary absence of RNS (MUL) from its cessation/loss (SNG).

Are there intermediate states of RNS?

The assembly of a functional RNS from ancestrally independent modules implies a stepwise process regardless of whether it occurred once (SNG) or multiple times (MUL), but MUL stretches the completion of the assembly long enough for lineages that initiated the process to have diverged from one another, such that the final developments of RNS are convergent (Fig. 2b). Under MUL incipient nodulators could possess intermediate conditions with some but not all modules found in nodulating species. Note that this is not necessarily the same as intermediate hidden states³, which need not be involved in nodulation. Under SNG, partially assembled RNS would represent decaying symbioses. In such intermediate states, NIN might retain interactions with genes from a subset of modules.

Conclusions

How novel processes and structures originate is a key question in evolutionary biology, and RNS and the nodule organ are an excellent plant system for addressing it. Kates et al.³ have produced a strong formulation of the multiple origins hypothesis that clarifies and can guide efforts to test the convergence theory. Notably, given the importance of legumes, MUL sensu Kates et al.³ identifies six origins and ten losses in the papilionoid + caesalpinioid clade. The rich diversity of RNS in crown papilionoids alone, homologous under either hypothesis, defies simple characterization, and much can be learned about how RNS evolves without straying from that single clade.

In our opinion, SNG¹⁵ remains a viable alternative to MUL, one that should be considered when evaluating RNS traits across diverse nodulating lineages. The fact that cessation is not loss blunts the conclusion that massive loss is highly improbable, and until non-RNS roles are identified for NIN and other genes lost or pseudogenized in non-nodulating NFNC species, the explanation provided by Kates et al.³ will remain unsatisfying.

For those pursuing the goal of engineering nitrogen fixation in crops outside the NFNC, a research program aimed at testing the competing hypotheses will be of great value simply by filling in the many gaps in our understanding of the diversity of nodulation symbioses. Whether achieving this goal is more vs. less likely under one or the other hypothesis is arguable, as Kates et al.³ correctly point out. As noted by Zhang et al.⁹, under either hypothesis, common features of nodulation in different lineages represent a likely minimum or core set of functions either retained (SNG) or independently assembled (MUL) for nodulation, with variable features being lineage-specific alternative innovations (MUL) or optimizations (SNG). In 1995, Soltis et al.⁴ charted the path forward to achieve both the fundamental and applied goals: Elucidate the biology of nodulating as well as non-nodulating NFNC species. On this, both Kates et al.³ and we agree.

References

Roy, P., Achom, M., Wilkinson, H., Lagunas, B. & Gifford, M. L. Symbiotic outcome modified by the diversification from 7 to over 700 nodule-specific cysteine-rich peptides. Genes 11, 348 (2020).
Article CAS PubMed PubMed Central Google Scholar
Guo, K., Yang, J., Yu, N., Luo, L. & Wang, E. Biological nitrogen fixation in cereal crops: progress, strategies, and perspectives. Plant Commun. 4, 100499 (2023).
Article CAS PubMed Google Scholar
Kates, H. R. et al. Shifts in evolutionary lability underlie independent gains and losses of root-nodule symbiosis in a single clade of plants. Nat. Commun. 15, 4262 (2024).
Article ADS CAS PubMed PubMed Central Google Scholar
Soltis, D. E. et al. Chloroplast gene sequence data suggest a single origin of the predisposition for symbiotic nitrogen fixation in angiosperms. Proc. Natl. Acad. Sci. USA 92, 2647–2651 (1995).
Article ADS CAS PubMed PubMed Central Google Scholar
Li, H. L. et al. Large-scale phylogenetic analyses reveal multiple gains of actinorhizal nitrogen-fixing symbioses in angiosperms associated with climate change. Sci. Rep. 5, 14023 (2015).
Article ADS PubMed PubMed Central Google Scholar
Griesmann, M. et al. Phylogenomics reveals multiple losses of nitrogen-fixing root nodule symbiosis. Science 361, eaat1743 (2018).
Article PubMed Google Scholar
Parniske, M. Uptake of bacteria into living plant cells, the unifying and distinct feature of the nitrogen-fixing root nodule symbiosis. Curr. Opin. Plant Biol. 44, 164–174 (2018).
Article CAS PubMed Google Scholar
Zuntini, A. R. et al. Phylogenomics and the rise of the angiosperms. Nature 629, 843–850 (2024).
Article ADS CAS PubMed PubMed Central Google Scholar
Zhang, Y. et al. Comparative phylogenomics and phylotranscriptomics provide insights into the genetic complexity of nitrogen-fixing root-nodule symbiosis. Plant Commun. 5, 100671 (2024).
Article CAS PubMed Google Scholar
Werner, G. D., Cornwell, W. K., Sprent, J. I., Kattge, J. & Kiers, E. T. A single evolutionary innovation drives the deep evolution of symbiotic N₂-fixation in angiosperms. Nat. Commun. 5, 4087 (2014).
Article ADS CAS PubMed Google Scholar
Doyle, J. J. Phylogeny of the legume family: an approach to understanding the origins of nodulation. Annu. Rev. Ecol. Evol. 25, 325–349 (1994).
Article Google Scholar
Doyle, J. J. Chasing unicorns: nodulation origins and the paradox of novelty. Am. J. Bot. 103, 1865–1868 (2016).
Article PubMed Google Scholar
Delaux, P. M. et al. Comparative phylogenomics uncovers the impact of symbiotic associations on host genome evolution. PLoS Genet. 10, e1004487 (2014).
Article PubMed PubMed Central Google Scholar
van Velzen, R. et al. Comparative genomics of the nonlegume Parasponia reveals insights into evolution of nitrogen-fixing rhizobium symbioses. Proc. Natl. Acad. Sci. USA 115, E4700–E4709 (2018).
PubMed PubMed Central Google Scholar
van Velzen, R., Doyle, J. J. & Geurts, R. A resurrected scenario: single gain and massive loss of nitrogen-fixing nodulation. Trends Plant Sci. 24, 49–57 (2019).
Article PubMed Google Scholar
Streeter, J. & Wong, P. P. Inhibition of legume nodule formation and nitrogen fixation by nitrate. Crit. Rev. Plant Sci. 7, 1–24 (1988).
Article CAS Google Scholar
Regus, J. U. et al. Nitrogen deposition decreases the benefits of symbiosis in a native legume. Plant Soil 414, 159–170 (2017).
Article ADS CAS Google Scholar
Nishida, H. & Suzaki, T. Nitrate-mediated control of root nodule symbiosis. Curr. Opin. Plant Biol. 44, 129–136 (2018).
Article CAS PubMed Google Scholar
Dupin, S. E., Geurts, R. & Kiers, E. T. The non-legume Parasponia andersonii mediates the fitness of nitrogen-fixing rhizobial symbionts under high nitrogen conditions. Front. Plant Sci. 10, 1779 (2019).
Article PubMed Google Scholar
Ardley, J. & Sprent, J. Evolution and biogeography of actinorhizal plants and legumes: a comparison. J. Ecol. 109, 1098–1121 (2021).
Article Google Scholar
Zhang, Y. G., Pagani, M., Liu, Z., Bohaty, S. M. & Deconto, R. A 40-million-year history of atmospheric CO₂. Philos. Trans. A Math. Phys. Eng. Sci. 371, 20130096 (2013).
ADS PubMed Google Scholar
Christin, P. A. et al. Oligocene CO₂ decline promoted C₄ photosynthesis in grasses. Curr. Biol. 18, 37–43 (2008).
Article CAS PubMed Google Scholar
Kürschner, W. M., Kvacek, Z. & Dilcher, D. L. The impact of Miocene atmospheric carbon dioxide fluctuations on climate and the evolution of terrestrial ecosystems. Proc. Natl. Acad. Sci. USA 105, 449–453 (2008).
Article ADS PubMed PubMed Central Google Scholar
Taylor, B. N. & Menge, D. N. L. Light regulates tropical symbiotic nitrogen fixation more strongly than soil nitrogen. Nat. Plants 4, 655–661 (2018).
Article CAS PubMed Google Scholar
Chen, H. & Markham, J. Ancient CO₂ levels favor nitrogen fixing plants over a broader range of soil N compared to present. Sci. Rep. 11, 3038 (2021).
Article ADS CAS PubMed PubMed Central Google Scholar
Elliott, G. N. et al. Burkholderia spp. are the most competitive symbionts of Mimosa, particularly under N-limited conditions. Environ. Microbiol. 11, 762–778 (2009).
Article PubMed Google Scholar
Oyserman, B. O. et al. Extracting the GEMs: genotype, environment, and microbiome interactions shaping host phenotypes. Front. Microbiol. 11, 574053 (2021).
Article PubMed PubMed Central Google Scholar
Fagorzi, C. et al. Nonadditive transcriptomic signatures of genotype-by-genotype interactions during the initiation of plant-rhizobium symbiosis. mSystems 6, 10–1128 (2021).
Article Google Scholar
Stern, D. B. & Crandall, K. A. The evolution of gene expression underlying vision loss in cave animals. Mol. Biol. Evol. 35, 2005–2014 (2018).
Article CAS PubMed PubMed Central Google Scholar
Moran, R. L. et al. Selection-driven trait loss in independently evolved cavefish populations. Nat. Commun. 14, 2557 (2023).
Article ADS CAS PubMed PubMed Central Google Scholar
Mergaert, P., Kereszt, A. & Kondorosi, E. Gene expression in nitrogen-fixing symbiotic nodule cells in Medicago truncatula and other nodulating plants. Plant Cell 32, 42–68 (2020).
Article CAS PubMed Google Scholar
Beaulieu, J. M., O’Meara, B. C. & Donoghue, M. J. Identifying hidden rate changes in the evolution of a binary morphological character: the evolution of plant habit in campanulid angiosperms. Syst. Biol. 62, 725–737 (2013).
Article PubMed Google Scholar
Boyko, J. D. & Beaulieu, J. M. Generalized hidden Markov models for phylogenetic comparative datasets. Methods Ecol. Evol. 12, 468–478 (2021).
Article Google Scholar
Coddington, J. A., Agnarsson, I., Hamilton, C. A. & Bond, J. E. Spiders did not repeatedly gain, but repeatedly lost, foraging webs. PeerJ 7, e6703 (2019).
Article PubMed PubMed Central Google Scholar
Goloboff, P. A., De Laet, J., Ríos-Tamayo, D. & Szumik, C. A. A reconsideration of inapplicable characters, and an approximation with step-matrix recoding. Cladistics 37, 596–629 (2021).
Article PubMed Google Scholar
Battenberg, K., Potter, D., Tabuloc, C. A., Chiu, J. C. & Berry, A. M. Comparative transcriptomic analysis of two actinorhizal plants and the legume Medicago truncatula supports the homology of root nodule symbioses and is congruent with a two-step process of evolution in the nitrogen-fixing clade of angiosperms. Front. Plant Sci. 9, 1256 (2018).
Article PubMed PubMed Central Google Scholar
Sprent, J. I. Which steps are essential for the formation of functional legume nodules? N. Phytol. 111, 129–153 (1989).
Article CAS Google Scholar
Nap, J. P. & Bisseling, T. Developmental biology of a plant-prokaryote symbiosis: the legume root nodule. Science 250, 948–954 (1990).
Article ADS CAS PubMed Google Scholar
Hirsch, A. M. & Larue, T. A. Is the legume nodule a modified root or stem or an organ sui generis? Crit. Rev. Plant Sci. 16, 361–392 (1997).
Article Google Scholar
Markmann, K. & Parniske, M. Evolution of root endosymbiosis with bacteria: how novel are nodules? Trends Plant Sci. 14, 77–86 (2009).
Article CAS PubMed Google Scholar
Soyano, T. & Hayashi, M. Transcriptional networks leading to symbiotic nodule organogenesis. Curr. Opin. Plant Biol. 20, 146–154 (2014).
Article CAS PubMed Google Scholar
Mathesius, U. Are legumes different? Origins and consequences of evolving nitrogen fixing symbioses. J. Plant Physiol. 276, 153765 (2022).
Article CAS PubMed Google Scholar
Doyle, J. J. Phylogenetic perspectives on the origins of nodulation. Mol. Plant Microbe Interact. 24, 1289–1295 (2011).
Article CAS PubMed Google Scholar
Shubin, N., Tabin, C. & Carroll, S. Deep homology and the origins of evolutionary novelty. Nature 457, 818–823 (2009).
Article ADS CAS PubMed Google Scholar
McKenna, K. Z., Wagner, G. P. & Cooper, K. L. A developmental perspective of homology and evolutionary novelty. Curr. Top. Dev. Biol. 141, 1–38 (2021).
Article PubMed Google Scholar
DiFrisco, J., Wagner, G. P. & Love, A. C. Reframing research on evolutionary novelty and co-option: character identity mechanisms versus deep homology. Semin. Cell Dev. Biol. 145, 3–12 (2023).
Article PubMed Google Scholar
Libourel, C. et al. Comparative phylotranscriptomics reveals ancestral and derived root nodule symbiosis programmes. Nat. Plants 9, 1067–1080 (2023).
Article CAS PubMed PubMed Central Google Scholar
Liu, S., Ratet, P. & Magne, K. Nodule diversity, evolution, organogenesis and identity. Adv. Bot. Res. 94, 119–148 (2020).
Article CAS Google Scholar
Corby, H. D. L. Types of rhizobial nodules and their distribution among the Leguminosae. Kirkia 13, 53–123 (1988).
Google Scholar
Shen, D. et al. A homeotic mutation changes legume nodule ontogeny into actinorhizal-type ontogeny. Plant Cell 32, 1868–1885 (2020).
Article CAS PubMed PubMed Central Google Scholar
Jhu, M. Y. & Oldroyd, G. E. D. Dancing to a different tune, can we switch from chemical to biological nitrogen fixation for sustainable food security? PLoS Biol. 21, e3001982 (2023).
Article CAS PubMed PubMed Central Google Scholar
Wang, D., Dong, W., Murray, J. & Wang, E. Innovation and appropriation in mycorrhizal and rhizobial symbioses. Plant Cell 34, 1573–1599 (2022).
Article PubMed PubMed Central Google Scholar
Liu, J. & Bisseling, T. Evolution of NIN and NIN-like genes in relation to nodule symbiosis. Genes 11, 777 (2020).
Article CAS PubMed PubMed Central Google Scholar
Soyano, T., Shimoda, Y., Kawaguchi, M. & Hayashi, M. A shared gene drives lateral root development and root nodule symbiosis pathways in Lotus. Science 366, 1021–1023 (2019).
Article ADS CAS PubMed Google Scholar
Soyano, T., Liu, M., Kawaguchi, M. & Hayashi, M. Leguminous nodule symbiosis involves recruitment of factors contributing to lateral root development. Curr. Opin. Plant Biol. 59, 102000 (2021).
Article CAS PubMed Google Scholar
Shen, L. & Feng, J. NIN-at the heart of nitrogen-fixing nodule symbiosis. Front. Plant Sci. 14, 1284720 (2023).
Article PubMed Google Scholar
Geurts, R. & Huisman, R. Innovations in two genes kickstarted the evolution of nitrogen-fixing nodules. Curr. Opin. Plant Biol. 77, 102446 (2024).
Article CAS PubMed Google Scholar
Mu, X. & Luo, J. Evolutionary analyses of NIN-like proteins in plants and their roles in nitrate signaling. Cell Mol. Life Sci. 76, 3753–3764 (2019).
Article CAS PubMed PubMed Central Google Scholar
Hsin, K. T., Yang, T. G., Lee, Y. H. & Cheng, Y. S. Phylogenetic and structural analysis of NIN-like proteins with a type I/II PB1 ___domain that regulates oligomerization for nitrate response. Front. Plant Sci. 12, 672035 (2021).
Article PubMed PubMed Central Google Scholar
Konishi, M., Okitsu, T. & Yanagisawa, S. Nitrate-responsive NIN-like protein transcription factors perform unique and redundant roles in Arabidopsis. J. Exp. Bot. 72, 5735–5750 (2021).
Article CAS PubMed Google Scholar
Zhao, Y. et al. Nuclear phylotranscriptomics and phylogenomics support numerous polyploidization events and hypotheses for the evolution of rhizobial nitrogen-fixing symbiosis in Fabaceae. Mol. Plant 14, 748–773 (2021).
Article CAS PubMed Google Scholar
Singh, S., Katzer, K., Katzer, J., Cerri, M. & Parniske, M. CYCLOPS, a DNA-binding transcriptional activator, orchestrates symbiotic root nodule development. Cell Host Microbe 15, 139–152 (2014).
Article CAS PubMed Google Scholar
Akamatsu, A., Nagae, M. & Takeda, N. The CYCLOPS Response Element in the NIN promoter is important but not essential for infection thread formation during Lotus japonicus–rhizobia symbiosis. Mol. Plant Microbe Interact. 35, 650–658 (2022).
Article CAS PubMed Google Scholar
Cathebras, C. et al. A novel cis-element enabled bacterial uptake by plant cells. Preprint at bioRxiv https://doi.org/10.1101/2022.03.28.486070 (2022).
Irving, T. B. et al. An LCO-responsive homolog of NODULE INCEPTION positively regulates lateral root formation in Populus sp. Plant Physiol. 190, 1699–1714 (2022).
Article CAS PubMed PubMed Central Google Scholar
Clavijo, F. et al. The Casuarina NIN gene is transcriptionally activated throughout Frankia root infection as well as in response to bacterial diffusible signals. N. Phytol. 208, 887–903 (2015).
Article CAS Google Scholar
Cissoko, M. et al. Actinorhizal signaling molecules: Frankia root hair deforming factor shares properties with NIN inducing factor. Front. Plant Sci. 9, 1494 (2018).
Article PubMed PubMed Central Google Scholar
Bu, F. et al. Mutant analysis in the nonlegume Parasponia andersonii identifies NIN and NF-YA1 transcription factors as a core genetic network in nitrogen-fixing nodule symbioses. N. Phytol. 226, 541–554 (2020).
Article CAS Google Scholar
Battenberg, K. & Hayashi, M. Evolution of root nodule symbiosis: focusing on the transcriptional regulation from the genomic point of view. Plant Biotechnol. 39, 79–83 (2022).
Article CAS Google Scholar
Liu, C. W. et al. NIN acts as a network hub controlling a growth module required for rhizobial infection. Plant Physiol. 179, 1704–1722 (2019).
Article CAS PubMed PubMed Central Google Scholar
Jiang, S. et al. NIN-like protein transcription factors regulate leghemoglobin genes in legume nodules. Science 374, 625–628 (2021).
Article ADS CAS PubMed Google Scholar
Yu, H. et al. Conserved cis-elements enable NODULES WITH ACTIVATED DEFENSE1 regulation by NODULE INCEPTION during nodulation. Plant Cell 36, 4622–4636 (2024).
Article PubMed Google Scholar
Soyano, T., Kouchi, H., Hirota, A. & Hayashi, M. NODULE INCEPTION directly targets NF-Y subunit genes to regulate essential processes of root nodule development in Lotus japonicus. PLoS Genet. 9, e1003352 (2013).
Article CAS PubMed PubMed Central Google Scholar
Vernié, T. et al. The NIN transcription factor coordinates diverse nodulation programs in different tissues of the Medicago truncatula root. Plant Cell 27, 3410–3424 (2015).
Article PubMed PubMed Central Google Scholar
Li, X. et al. RPG interacts with E3-ligase CERBERUS to mediate rhizobial infection in Lotus japonicus. PLoS Genet. 19, e1010621 (2023).
Article CAS PubMed PubMed Central Google Scholar
Feng, J., Lee, T., Schiessl, K. & Oldroyd, G. E. D. Processing of NODULE INCEPTION controls the transition to nitrogen fixation in root nodules. Science 374, 629–632 (2021).
Article ADS CAS PubMed Google Scholar
Konishi, M. & Yanagisawa, S. Arabidopsis NIN-like transcription factors have a central role in nitrate signalling. Nat. Commun. 4, 1617 (2013).
Article ADS PubMed Google Scholar
Panchy, N., Lehti-Shiu, M. & Shiu, S. H. Evolution of gene duplication in plants. Plant Physiol. 17, 2294–2316 (2016).
Article Google Scholar
Fu, M., Sun, J., Li, X., Guan, Y. & Xie, F. Asymmetric redundancy of soybean Nodule Inception (NIN) genes in root nodule symbiosis. Plant Physiol. 188, 477–489 (2022).
Article CAS PubMed Google Scholar
Cervantes-Pérez, S. A. et al. Single-cell transcriptome atlases of soybean root and mature nodule reveal new regulatory programs that control the nodulation process. Plant Commun. 5, 100984 (2024).
Article PubMed PubMed Central Google Scholar
O’Malley, R. C. et al. Cistrome and epicistrome features shape the regulatory DNA landscape. Cell 165, 1280–1292 (2016).
Article PubMed PubMed Central Google Scholar
de Boer, C. G. & Taipale, J. Hold out the genome: a roadmap to solving the cis-regulatory code. Nature 625, 41–50 (2024).
Article ADS PubMed Google Scholar
Zhu, T. et al. Comprehensive mapping and modelling of the rice regulome landscape unveils the regulatory architecture underlying complex traits. Nat. Commun. 15, 6562 (2024).
Article CAS PubMed PubMed Central Google Scholar
Johnson, A. D. The rewiring of transcription circuits in evolution. Curr. Opin. Genet. Dev. 47, 121–127 (2017).
Article CAS PubMed PubMed Central Google Scholar
Schmitz, R. J., Grotewold, E. & Stam, M. Cis-regulatory sequences in plants: their importance, discovery, and future challenges. Plant Cell 34, 718–741 (2022).
Article PubMed Google Scholar
Pereira, W. J. et al. Functional and comparative genomics reveals conserved noncoding sequences in the nitrogen-fixing clade. N. Phytol. 234, 634–649 (2022).
Article CAS Google Scholar
Lace, B. et al. RPG acts as a central determinant for infectosome formation and cellular polarization during intracellular rhizobial infections. eLife 12, e80741 (2023).
Article CAS PubMed PubMed Central Google Scholar
Mendieta, J. P. et al. Investigating the cis-regulatory basis of C₃ and C₄ photosynthesis in grasses at single-cell resolution. Proc. Natl. Acad. Sci. USA 121, e2402781121 (2024).
Article CAS PubMed PubMed Central Google Scholar
Reed, J. et al. Elucidation of the pathway for biosynthesis of saponin adjuvants from the soapbark tree. Science 379, 1252–1264 (2023).
Article ADS CAS PubMed Google Scholar
Billault-Penneteau, B., Sandré, A., Folgmann, J., Parniske, M. & Pawlowski, K. Dryas as a model for studying the root symbioses of the Rosaceae. Front. Plant Sci. 10, 661 (2019).
Article PubMed PubMed Central Google Scholar
Rojas-Jimenez, K. et al. Exploring symbiotic legume-rhizobia relationships across tropical species. Preprint at bioRxiv https://doi.org/10.1101/2024.08.23.609460 (2024).
O’Dell, R. E., James, J. J. & Richards, J. H. Congeneric serpentine and nonserpentine shrubs differ more in leaf Ca: Mg than in tolerance of low N, low P, or heavy metals. Plant Soil 280, 49–64 (2006).
Article Google Scholar
Terra, V. et al. Dilemmas in generic delimitation of Senegalia and allies (Caesalpinioideae, mimosoid clade): how to reconcile phylogenomic evidence with morphology and taxonomy? PhytoKeys 205, 261–278 (2022).
Article PubMed PubMed Central Google Scholar
Harrier, L. A., Whitty, P. W., Sutherland, J. M. & Sprent, J. I. Phenetic investigation of non-nodulating African species of Acacia (Leguminosae) using morphological and molecular markers. Plant Syst. Evol. 205, 27–51 (1997).
Article Google Scholar
Sequerra, J., Capellano, A., Gianinazzi-Pearson, V. & Moiroud, A. Ultrastructure of cortical root cells of Alnus incana infected by Penicillium nodositatum. N. Phytol. 130, 545–555 (1995).
Article Google Scholar
Béreau, M. & Garbaye, J. First observations on the root morphology and symbioses of 21 major tree species in the primary tropical rain forest of French Guyana. Ann. For. Sci. 51, 407–416 (1994).
Article Google Scholar
Duhoux, E. et al. Angiosperm Gymnostoma trees produce root nodules colonized by arbuscular mycorrhizal fungi related to Glomus. N. Phytol. 149, 115–125 (2001).
Article CAS Google Scholar
Rahimlou, S. et al. Unveiling root nodulation in Tribulus terrestris and Roystonea regia via metagenomics analysis. Mol. Genet. Genomics 300, 9 (2024).
Article PubMed Google Scholar
Farris, J. S. Phylogenetic analysis under Dollo’s Law. Syst. Biol. 26, 77–88 (1977).
Article Google Scholar
Zanne, A. E. et al. Three keys to the radiation of angiosperms into freezing environments. Nature 506, 89–92 (2014).
Article ADS CAS PubMed Google Scholar
Cannon, S. B. et al. Multiple polyploidy events in the early radiation of nodulating and nonnodulating legumes. Mol. Biol. Evol. 32, 193–210 (2015).
Article CAS PubMed Google Scholar
Azani, N. A new subfamily classification of the leguminosae based on a taxonomically comprehensive phylogeny: the legume phylogeny working group (LPWG). Taxon 66, 44–77 (2017).
Article Google Scholar
Edwards, E. J. Evolutionary trajectories, accessibility and other metaphors: the case of C₄ and CAM photosynthesis. N. Phytol. 223, 1742–1755 (2019).
Article Google Scholar
Edwards, E. J. Reconciling continuous and discrete models of C₄ and CAM evolution. Ann. Bot. 132, 717–725 (2023).
Article CAS PubMed PubMed Central Google Scholar
Sage, R. F., Christin, P. A. & Edwards, E. J. The C₄ plant lineages of planet Earth. J. Exp. Bot. 62, 3155–3169 (2011).
Article CAS PubMed Google Scholar
Swift, J. et al. Exaptation of ancestral cell-identity networks enables C₄ photosynthesis. Nature 636, 143–150 (2024).
Article CAS PubMed PubMed Central Google Scholar
Christin, P.-A., Freckleton, R. P. & Osborne, C. P. Can phylogenetics identify C₄ origins and reversals? Trends Ecol. Evol 25, 403–409 (2010).
Article PubMed Google Scholar
Leebens-Mack, J. et al. One thousand plant transcriptomes and the phylogenomics of green plants. Nature 574, 679–685 (2019).
Article Google Scholar
Thomas, B. C., Pedersen, B. & Freeling, M. Following tetraploidy in an Arabidopsis ancestor, genes were removed preferentially from one homeolog leaving clusters enriched in dose-sensitive genes. Genome Res. 16, 934–946 (2006).
Article CAS PubMed PubMed Central Google Scholar
Op den Camp, R. H. M. et al. A phylogenetic strategy based on a legume-specific whole genome duplication yields symbiotic cytokinin type-A response regulators. Plant Physiol. 157, 2013–2022 (2011).
Article CAS PubMed PubMed Central Google Scholar
Young, N. D. et al. The Medicago genome provides insight into the evolution of rhizobial symbioses. Nature 480, 520–524 (2011).
Article ADS CAS PubMed PubMed Central Google Scholar
Cannon, S. B. et al. Polyploidy did not predate the evolution of nodulation in all legumes. PLoS ONE 5, e11630 (2010).
Article ADS PubMed PubMed Central Google Scholar
Li, Q. G., Zhang, L., Li, C., Dunwell, J. M. & Zhang, Y. M. Comparative genomics suggests that an ancestral polyploidy event leads to enhanced root nodule symbiosis in the Papilionoideae. Mol. Biol. Evol. 30, 2602–2611 (2013).
Article CAS PubMed Google Scholar
Czyż, K. B., Taylor, C. M., Kawaliło, M. & Koczyk, G. Gain or loss? Evidence for legume predisposition to symbiotic interactions with rhizobia via loss of pathogen-resistance-related gene families. Int. J. Mol. Sci. 23, 16003 (2022).
Article PubMed PubMed Central Google Scholar
Powell, A. F. & Doyle, J. J. Enhanced rhizobial symbiotic capacity in an allopolyploid species of Glycine (Leguminosae). Am. J. Bot. 103, 1771–1782 (2016).
Article PubMed Google Scholar
Forrester, N. J. & Ashman, T. L. Autopolyploidy alters nodule-level interactions in the legume-rhizobium mutualism. Am. J. Bot. 107, 179–185 (2020).
Article CAS PubMed Google Scholar
Koenen, E. J. M. et al. The origin of the legumes is a complex paleopolyploid phylogenomic tangle closely associated with the Cretaceous-Paleogene (K-Pg) mass extinction event. Syst. Biol. 70, 508–526 (2021).
Article PubMed Google Scholar
Wang, J. et al. A common whole-genome paleotetraploidization in Cucurbitales. Plant Physiol. 190, 2430–2448 (2022).
Article CAS PubMed PubMed Central Google Scholar
Wendel, J. F. The wondrous cycles of polyploidy in plants. Am. J. Bot. 102, 1753–1756 (2015).
Article CAS PubMed Google Scholar
Parshuram, Z. A., Harrison, T. L., Simonsen, A. K., Stinchcombe, J. R. & Frederickson, M. E. Nonsymbiotic legumes are more invasive, but only if polyploid. N. Phytol. 237, 758–765 (2023).
Article CAS Google Scholar
Powell, A. F. & Doyle, J. J. 2015. The implications of polyploidy for the evolution of signalling in rhizobial nodulation symbiosis. Adv. Bot. Res. 75, 149–190 (2015).
Article Google Scholar
Moharana, K. C. & Venancio, T. M. Polyploidization events shaped the transcription factor repertoires in legumes (Fabaceae). Plant J. 103, 726–741 (2020).
Article CAS PubMed Google Scholar
Liu, T. et al. Duplication and sub-functionalization of flavonoid biosynthesis genes plays important role in Leguminosae root nodule symbiosis evolution. J. Integr. Plant Biol. 66, 2191–2207 (2024).
Article CAS PubMed Google Scholar

Download references

Acknowledgements

Makoto Hayashi, Martin Parniske, and Rene Geurts kindly answered questions about their work. Jesus Martinez Gomez and Mick Song answered questions about hidden rate modeling. We appreciate two insightful reviews, provided by Erika Edwards and by Steven Cannon, Jacob Stai, and Hyunoh Lee. Colin Hughes kindly read and commented on the revised manuscript. We are grateful to Ryan Folk for clarifying aspects of the Kates et al. manuscript.

Author information

These authors contributed equally: Jeff J. Doyle, Jian Ren.

Authors and Affiliations

School of Integrative Plant Science, Section of Plant Biology and Section of Plant Breeding & Genetics, Cornell University, Ithaca, NY, USA
Jeff J. Doyle
Institute of Grassland Science, Key Laboratory of Vegetation Ecology of the Ministry of Education, Jilin Songnen Grassland Ecosystem National Observation and Research Station, State Environmental Protection Key Laboratory of Wetland Ecology and Vegetation Restoration, Northeast Normal University, Changchun, China
Jian Ren & Yingzhi Gao
Jilin Academy of Agricultural Sciences/National Engineering Research Center for Soybean, Changchun, Jilin, China
Jian Ren
Department of Ecology, Environment and Plant Science, Stockholm University, Stockholm, Sweden
Katharina Pawlowski
The James Hutton Institute, Dundee, UK
Euan K. James
Key Laboratory of Grassland Resources and Ecology of Western Arid Desert Area of the Ministry of Education, College of Grassland Science, Xinjiang Agricultural University, Urumqi, China
Yingzhi Gao

Authors

Jeff J. Doyle
View author publications
Search author on:PubMed Google Scholar
Jian Ren
View author publications
Search author on:PubMed Google Scholar
Katharina Pawlowski
View author publications
Search author on:PubMed Google Scholar
Euan K. James
View author publications
Search author on:PubMed Google Scholar
Yingzhi Gao
View author publications
Search author on:PubMed Google Scholar

Contributions

J.R. and Y.G. conceived the overall project. J.D. and J.R. wrote the manuscript. All authors contributed intellectually to the project.

Corresponding authors

Correspondence to Jeff J. Doyle or Yingzhi Gao.

Ethics declarations

Competing interests

The authors declare no competing interests.

Peer review

Peer review information

Nature Communications thanks Steven Cannon, who co-reviewed with Hyunoh Lee and Jacob Stai, and Erika Edwards for their contribution to the peer review of this work. A peer review file is available.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.

Reprints and permissions

About this article

Cite this article

Doyle, J.J., Ren, J., Pawlowski, K. et al. One versus many independent assemblies of symbiotic nitrogen fixation in flowering plants. Nat Commun 16, 5345 (2025). https://doi.org/10.1038/s41467-025-60433-w

Download citation

Received: 04 November 2024
Accepted: 23 May 2025
Published: 19 June 2025
DOI: https://doi.org/10.1038/s41467-025-60433-w